Automated Repricers and the Buy Box: When Bots Set Your Price

The Buy Box moves faster than any person can

On a shared listing, the Buy Box decides who makes the sale, and it does not sit still. Prices on busy listings change all day as sellers come and go, and the winner rotates many times a day. A seller with a few dozen listings can keep up by hand. A seller with a few hundred cannot, which is why repricing software exists. It connects through the marketplace API, pulls your listings and costs, watches the current Buy Box price and the lowest competing offers, and submits a new price the moment a trigger fires. Amazon's own Automate Pricing tool processes most updates in under fifteen minutes. Third-party tools that connect through the same API claim response times around one second on their fastest tiers. Either way, the race is measured in minutes or seconds, not days.

There are two kinds of repricer, and the difference matters more than the price of the tool. A rule-based repricer follows instructions you write: match the Buy Box, stay a cent under the lowest competing offer, or hold a set margin. An AI repricer tests moves and learns which ones win the Buy Box, which can protect margins on busy listings because it sometimes learns that a slightly higher price still wins. Both are legitimate tools. Both become a problem the same way, when the rules have no floor.

The price war nobody agreed to

Put two rule-based repricers on the same listing, each told to stay a cent under the other, and you get what sellers call price ping-pong. Each tool sees the competitor drop, drops a cent lower, and waits. The other tool does the same. Prices fall in tiny steps, several times an hour, until one of them hits a floor or a seller notices the margin is gone. The Informed Repricer guide names the missing guardrails as the most common configuration mistake in repricing: without a floor, a price war between two competing repricers can drive prices toward zero through rapid oscillation, and without a ceiling, prices never recover when competition thins out.

This is not price fixing in the legal sense. The Federal Trade Commission defines price fixing as an agreement among competitors to raise, lower, maintain, or stabilize prices. Two repricers chasing each other down is the opposite: no agreement, no communication, just two machines following rules their owners wrote separately. That makes it harder to blame and harder to stop, because neither seller chose a price war. Each one chose a rule, and the rules collided.

Build the floor from real numbers

The floor is the one number that decides whether automation helps you or quietly empties your margin. Build it from costs, not from the price you hope to get: cost of goods, fulfillment fee, referral fee, shipping, and the margin you actually need. If that floor is above the current Buy Box price, the right move is to lose the Buy Box for a while, not to chase it below cost. A repricer that cannot lose is a repricer that will sell at a loss, and the loss will not show up as a line item. It shows up as a month where sales rose and profit fell, spread across hundreds of orders as a slightly lower average selling price.

The second guardrail is a ceiling and a reset rule. When a competitor sells out or leaves the listing, your price should climb back toward your target, not sit at the bottom of a war that ended hours ago. And the third is a pause rule for the listings that matter most: if a repricer has moved the same SKU more than a set number of times in a day, something is wrong, and a person should look at it before the next move. Automation earns its keep on the listings you do not have time to watch. It earns nothing on the listings where it is fighting a mirror image of itself. If a hijacked account is the bigger risk on your storefront, read how a hijacked seller account sells the attacker's goods on your reputation, and if the listings themselves look wrong, scraper storefronts and ghost inventory show what a listing with no seller behind it looks like.